Most business intelligence tools tell you what happened. The session Divya Manohar ran at Epicor Insights 2026 was about what to do next. That distinction matters more than any feature list.

Manohar, an account manager at Phocas Software who spent her first years implementing the product before moving into sales, has been with the company for a decade. She knows where the gaps are. The session, titled Drive Sales Growth with Epicor Data Analytics, Powered by Phocas Software, walked a room of Epicor customers through a practical demonstration: how do you turn a mandate to grow sales by 10 percent this quarter into a set of specific, data-backed actions a sales rep can actually execute?

The answer Phocas is building toward involves three layers working in sequence. Analytical intelligence surfaces what the data says. Predictive logic identifies where risk is forming. Generative output converts both into next steps a rep can act on immediately. The live demo showed all three in motion, and the architecture behind it is worth examining closely.

The constraint Phocas is solving

Epicor customers in distribution and manufacturing are not short on data. They have sales history, purchasing records, inventory positions, margin detail by sales representative, and rebate tracking. The problem is accessibility. Most of that data lives inside the enterprise resource planning system, readable only by people who know how to query it. Phocas sits above the enterprise resource planning layer, pulling data through pre-built connectors for Kinetic, Prophet 21, Eclipse, and other Epicor platforms via a sync client that handles the extraction nightly, or hourly for customers who need it.

The interface is designed for business users, not database administrators. Two clicks to filter a customer list to accounts above $20,000 in sales with margins below a threshold. One toggle to switch from raw numbers to visual totals. Direct drill-down from a territory view to individual SKUs showing negative margin. The technical architecture is not the story. The adoption number is: Phocas claims 50 percent higher user adoption rates compared to other business analytics tools in its category. That is a credibility claim worth pressure-testing, but the product design philosophy behind it is visible in the demo. Complexity is hidden; decisions are surfaced.

What RFM segmentation actually reveals

The session introduced the recency, frequency, monetary model as the conceptual backbone of EDA Sales Insights. The model is not new. What Phocas has done is operationalize it directly against Epicor customer data and make the output navigable without a data team.

The segmentation chart groups customers into ten categories ranging from Champions at one end, customers who buy frequently, spend the most, and bought most recently, to Hibernating at the other, customers who have not engaged in a significant period. The live data in the demo showed 133 Champion customers generating more than half of total revenue across 770 accounts. The practical implication is blunt: most sales teams are spending most of their time with the accounts that need the least attention, while the accounts at risk of churning receive little structured outreach.

The segmentation thresholds are adjustable. A distributor whose sales territory handles large industrial accounts will define "high frequency" differently than one managing a high-volume retail base. Phocas lets administrators set the recency, frequency, and monetary cutoffs by territory or sales representative, so the segmentation reflects actual business context rather than a generic algorithm.

The AI layer: three modes, one platform

EDA Sales Insights launched in October 2025 and already has more than 150 customers in its early access program. The AI functionality embedded in the customer profile view is where the product moves from reporting to recommendation.

When a sales representative clicks into a customer profile flagged as At Risk or Needs Attention, the platform generates an AI summary pulling from the Phocas data directly. The demo showed a profile for a customer classified as a Champion where three products that collectively drove over $55,000 in the prior period had dropped to zero in the current one. Revenue overall was up 10 percent, which would mask the churn signal in a standard report. The AI summary surfaced it, named the specific products, and generated a set of next steps: ask the customer directly whether they switched supplier, protect the remaining volume before it follows the same trajectory, and bring a value story rather than a price concession given the account's 6 percent margin profile.

That is analytical AI reading the data, predictive logic identifying the risk pattern, and generative output converting both into a specific action sequence. The AI component comes included with the Phocas platform. It is not a separately licensed add-on. That pricing decision is itself a signal: Phocas is betting that AI-assisted guidance drives retention and expansion revenue, making it worth embedding in the base product rather than extracting additional margin from it.

The platform also includes a natural language query interface. A sales manager can type a question directly into the Phocas AI interface and receive a generated chart or table. The session demonstrated a query for the top ten sales representatives by last quarter performance, rendered as a visual in seconds. For organizations where getting a custom report previously required an IT ticket, this changes the access model entirely.

CRM integration and the task loop

Identifying a customer at risk is only useful if the insight generates action. Phocas closes that loop through a CRM product that operates alongside Sales Insights. From the segmentation view, a manager can select a group of at-risk accounts and create tasks assigned directly to the sales representatives responsible for those accounts. The task record shows the customer's last purchase date, total order count, and lifetime spend. The representative sees it in the Phocas agenda, which connects to Microsoft Outlook.

The mobile view of the Phocas CRM allows a sales representative to call, text, or email a customer directly from the account record, and to log a call note via voice input that the system transcribes and saves as an activity. For field sales teams that have historically resisted CRM adoption because of data entry friction, this is a credible answer to a real objection.

The question from the room about whether Phocas can predict when a customer is about to slide from About to Sleep into At Risk status, before the transition happens, drew a candid response: that capability is in development. The framing was a predictive alert that surfaces a flag 60 days before a customer's buying pattern deteriorates to the threshold. It is not available yet, but the product team is actively building toward it.

The product vision and where it is heading

The roadmap slide shown at the session frames Phocas as a platform covering three functional domains above the enterprise resource planning layer: Sales, covering sales insights, CRM, and sales forecasting; Finance, covering financial statements, budgets and forecasts, workforce planning, and board reports; and Operations, covering rebates management, operational forecasting, and inventory planning. The connective layer underneath all of it is described as Data You Can Trust, combining ERP integrations, industry expertise, and large language model connectivity.

That last phrase deserves attention. Phocas has released a Model Context Protocol server, which means the platform's data can be surfaced inside tools like Microsoft Copilot or other large language model interfaces. The session presenter was careful not to overpromise on what that enables today, but the architectural implication is clear: Phocas is positioning its curated, industry-specific data layer as the trusted input for whatever AI interface a customer's organization is already using.

Inventory predictive analysis is also on the roadmap, with the stated goal of flagging dead stock, low inventory risk, and reorder recommendations based on purchasing trends. The product team is in active customer interviews to define what that capability should look like in practice.

The ecosystem position and the migration question

As noted in the earlier coverage of the Epicor Insights 2026 keynote on this site, Phocas is one of several ISV partners that represent a meaningful migration dependency for Epicor customers. A distributor running Phocas for analytics, budgets, and CRM is not switching enterprise resource planning systems on a short timeline. The integration runs at the data layer, not just the surface. Phocas's value to Epicor's customer retention strategy is proportional to how deeply embedded its platform becomes in day-to-day sales and finance workflows.

The product design philosophy reinforces this. Phocas built pre-configured connectors for every major Epicor platform. It built an onboarding path that gets customers running on sales data without requiring database expertise. It priced AI assistance into the base product rather than reserving it for higher tiers. Each of those decisions increases switching cost and daily usage frequency simultaneously.